Direct in-treatment imaging of lung tumors during radiation therapy to ensure proper radiation targeting is an unsolved challenge. We propose to overcome this obstacle using beam's-eye-view (BEV) imaging with a high efficiency imager to automatically localize lung tumors in real-time. The project goal is to design, validate and clinically deploy the new high efficiency imager. We will first determine the specifications necessary for robust BEV imaging. The design of the novel imager to meet this goal will be performed using a new Monte Carlo platform. Scintillator composition, pixelation and stacking of multiple layers will be optimized and the new imager will be validated prior to full scale deployment in a radiation therapy clinic. In addition to the immediate clinical benefits of real-tie lung tumor localization, we anticipate that this project will accelerate advanced applications like tumor tracking, delivered dose accumulation and adaptive radiation therapy. The Monte Carlo platform will also hasten the development of novel imagers for other medical applications beyond this project. Therefore the impact of the project extends far beyond the immediate clinical gain, which is in itself substantial.